Re: Split Issue 30?

Silvia Pfeiffer, Mon, 13 Feb 2012 12:22:51 +1100:
> On Mon, Feb 13, 2012 at 11:19 AM, Charles Pritchard <> wrote:
>> On 2/12/2012 3:42 PM, Jonas Sicking wrote:

>> My comment was intended as: we should wait to break current behavior, until
>> HTML6.
> You keep saying that it will beak current behaviour. But that's not
> the case. The @longdesc is only deprecated, it is not removed from
> HTML5. All browsers will continue to support it. The only difference
> will be that validators will flag it as deprecated and can even
> provide information as to what people should do instead. I believe in
> such a mechanism as a productive way of moving authors forward into
> using better features, while at the same time continuing to support
> old features such as @longdesc. If we all really believe that there
> are better mechanisms than @longdesc, then we should not dwell on it
> but instead make use of this constructive way of moving towards a
> better future.

Steve's web site does not list support for 
@aria-describedby at all - I guess he has not tested it in a systematic 
way. [Take it as a hint, Steve ...] 

I should have had the hard facts on support level in front of me, but I 
don't. However, from previous and current testing, it seems to be hard 
to find AT that support @aria-describedby - at all. In the @summary 
attribute case, I have played in the same message: @Summary works. But 
AT currently does not react to the ARIA alternatives, if you try to use 
them instead of @summary. For <img>, then 
@aria-describedby/@aria-labelledby works slightly better than with any 
other element, but even there it is far from perfect.

Most users are still on Windows XP I believe. And all browser vendors 
try to be XP-compatible. I tried very hard, tonight, to make 
@aria-describedby work on XP for the <img> element together with NVDA: 
Mission unaccomplished.

If we just tell authors to use @aria-describedby - and ditch @longdesc 
- even when ARIA support is at the current level, what do we have then, 
if not a cargo cult?

It is nice and dandy that @longdesc is required to work - even if it is 
not valid. But do you really expect e.g. VoiceOver to support @longdesc 
if it isn't considered valid? I would like to believe it, but I don't. 
Has @longdesc been added to the test suite? Could I do that if it 
hasn't? Would it be accepted?
Leif Halvard Silli

Received on Monday, 13 February 2012 02:10:20 UTC